Back to articles
Volume: 32 | Article ID: art00014
An Evaluation of Embedded GPU Systems for Visual SLAM Algorithms
  DOI :  10.2352/ISSN.2470-1173.2020.6.IRIACV-325  Published OnlineJanuary 2020

Simultaneous Localization and Mapping (SLAM) solves the computational problem of estimating the location of a robot and the map of the environment. SLAM is widely used in the area of navigation, odometry, and mobile robot mapping. However, the performance and efficiency of the small industrial mobile robots and unmanned aerial vehicles (UAVs) are highly constrained to the battery capacity. Therefore, a mobile robot, especially a UAV, requires low power consumption while maintaining high performance. This paper demonstrates holistic and quantitative performance evaluations of embedded computing devices that run on the Nvidia Jetson platform. Evaluations are based on the execution of two state-of-the-art Visual SLAM algorithms, ORB-SLAM2 and OpenVSLAM, on Nvidia Jetson Nano, Nvidia Jetson TX2, and Nvidia Jetson Xavier.

Subject Areas :
Views 46
Downloads 9
 articleview.views 46
 articleview.downloads 9
  Cite this article 

Tao Peng, Dingnan Zhang, Don Lahiru Nirmal Hettiarachchi, John Loomis, "An Evaluation of Embedded GPU Systems for Visual SLAM Algorithmsin Proc. IS&T Int’l. Symp. on Electronic Imaging: Intelligent Robotics and Industrial Applications using Computer Vision,  2020,  pp 325-1 - 325-6,

 Copy citation
  Copyright statement 
Copyright © Society for Imaging Science and Technology 2020
Electronic Imaging
Society for Imaging Science and Technology
7003 Kilworth Lane, Springfield, VA 22151 USA